A Fast and Robust Regression Model for Keyframes Extraction

In recent days, multimedia content has been developed from multiple resources such as social media video content, video surveillance, etc. Due to rapid increasing the volume of digital data, analysis and retrieval of the video will be time consuming. Due to continuous frame count, retrieving content of the video from large databases is a challenging one. To overcome the above challenge, this paper introduces a regression model to extract the information that is derived from frame sequence that uses two statistical features on the frame. This model can be applied on any kinds of video data to extend several multimedia applications and experiments for scene understanding, video summarization, etc Finally, user generated video where static video and dynamic videos and VIRAT benchmark dataset are evaluated by demonstrating the effectiveness of the proposed work.